Thursday, December 6, 2012

AND it's a.........

So last night before we went in for our ultrasound, we decided to try a few different gender prediction tests. We're curious and empirical, so we're going to give you the low down on each of them. Then you, as the educated reader, can decide which ones are most accurate.

The Tests

1. How the baby is carried (high=girl, low=boy): BOY
2. Heartbeat (under 140=boy, above 140=girl): GIRL
3. Cravings (sweet=girl, salty/sour=boy): GIRL
4. Ancient Chinese Prediction Method (Age at conception + Month of conception): GIRL
5. Ancient Mayan Prediction Method (Age and Year of conception are both even or odd=girl, not the same=boy): GIRL
6. How a key is picked up (narrow part=girl, round part=boy): GIRL
7. Pregnancy acne (normal=boy, worse=girl): BOY
8. Swinging wedding ring (circle=boy, back and forth=girl): BOY
9. Morning sickness in first trimester (bad=girl, little or none=boy): GIRL
10. How baby weight is carried (hips=girl, tummy=boy): GIRL

Our Analysis

We decided to use a weighted average of our ten-dimensional vector to statistically combine the results of all ten tests. The weights are a number between 1 and 0 that conveys how much we trust that particular measure (1 = perfect trust, 0 = no trust whatsoever). The vector of predictions (Vp) is first assembled thus (1 = girl, -1 = boy):

[-1, 1, 1, 1, 1, 1, -1, -1, 1, 1]

 Our trust values are chosen based on the following reasons:
1. We don't trust this as much because in Kangeroos it can be misleading. Trust = 0.4
2. We trust this measure based on average laziness of genders as measured by conference talks about the need to work harder (although the measure loses predictive power during puberty when in the presence of the opposite gender). Trust = 0.8
3. Jessica always wants sweet stuff. How predictive could that be? Trust = 0.3
4,5. Ancient methods agree, and are tried and true. Trust = 0.95433234
6. Based on solid physical principles. Trust = 0.8
7. Results of this predictor can be confounded with craving predictor, and they don't agree. Trust = 0.45
 8. I've seen big pendulums in science museums predicting the baby's gender inside mother earth, and I trust the curators of science museums generally. Trust = 0.9
9. Girls don't make me barfy. This is simply absurd. Trust = 0.01
10. No comment on baby weight from the analyst. Trust is a neutral 0.5

Trust vector (Vt):

[0.4, 0.8, 0.3, 0.95433234, 0.95433234, 0.8, 0.45, 0.9, 0.01, 0.5]

Final weighted average =
Vp * Vt / 10 = 0.2569
0.2569 > 0

It might be a GIRL. That's the official prediction from Matt "Nate Silver" and Jessica "Nostradamus".












The Real Deal

Ok, so after the fun with at-home prediction methods, we wanted to double check our answer. We went in to do the ultrasound and got a pretty clear result. You can see for yourself!


 And another adorable shot of our little boy sucking his thumb! We're in love already!


Works Cited:
1. M. R. Kosorok et al. 19something
2. Nate Silver on Bubbles, 20something
3. Nostradamus et al. The Prophecies, 15something

4 comments:

  1. I am so excited for you! I can't wait!

    ReplyDelete
  2. One of the BEST posts I have ever read! Oh. and by the way . . . CONGRATULATIONS!

    ReplyDelete
  3. SO funny!! LOVE IT!! Congratulations Mom & Dad!! We are SO VERY excited for you two!

    ReplyDelete